「alignment-testing」の検索結果

Systematically bypass LLM safety filters and content moderation systems using advanced jailbreaking techniques. Use this skill when testing AI systems for safety alignment robustness, evaluating content moderation effectiveness, or conducting authorized AI red team assessments. Covers role-play attacks, few-shot manipulation, encoding bypass, multi-modal exploitation, and automated jailbreak discovery methods.

日本語の概要は準備中です。原文の説明を表示しています。